Research map / Label-efficient and robust learning / Test-time adaptation
Point-TTA: Test-Time Adaptation for Point Cloud Registration Using Multitask Meta-Auxiliary Learning
Ahmed Hatem, Yiming Qian, Yang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add meta-auxiliary test-time adaptation framework | DGR · 3DMatch | Registration Recall: 91.30 -> 92.45 | Improved |
| add auxiliary tasks only (no meta, no TTA) (ablation) | DGR · 3DMatch | Registration Recall: 91.31 -> 91.42 | Improved |
| add auxiliary tasks only (no meta, no TTA) (ablation) | DGR · KITTI (trained on 3DMatch) | Registration Recall: 95.24 -> 95.21 | No clear effect |
| add test-time adaptation to auxiliary learning (ablation) | DGR · 3DMatch | Registration Recall: 91.42 -> 91.86 | Improved |
| add meta-auxiliary training (ablation) | DGR · 3DMatch | Registration Recall: 91.86 -> 92.28 | Improved |
| add point cloud reconstruction auxiliary task (ablation) | DGR · 3DMatch | Registration Recall: 91.31 -> 92.24 | Improved |
| add BYOL feature learning task on top of rec and cc (ablation) | DGR · 3DMatch | Rotation Error (deg): 1.69 -> 1.71 | Mixed |
| use learnable balancing weights instead of fixed (ablation) | DGR · 3DMatch | Registration Recall: 92.26 -> 92.45 | Improved |
| apply method to cross-dataset ETH registration | DGR · ETH | Average Recall: 81.78 -> 86.85 | Improved |
| apply method on low-overlap 3DLoMatch | DGR · 3DLoMatch | Registration Recall: 43.80 -> 50.73 | Improved |
Closest papers by meaning
- Deep Learning-Based Point Cloud Registration: A Comprehensive Survey and Taxonomy (IJCV 2026 · similarity 0.91)
- SIRA-PCR: Sim-to-Real Adaptation for 3D Point Cloud Registration (ICCV 2023 · similarity 0.90)
- SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds (ICML 2025 · similarity 0.87)
- MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing (CVPR 2025 · similarity 0.86)
- PointMAC: Meta-Learned Adaptation for Robust Test-Time Point Cloud Completion (NeurIPS 2025 · similarity 0.86)
- PCoTTA: Continual Test-Time Adaptation for Multi-Task Point Cloud Understanding (NeurIPS 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
